Jeremy Wacksman

We've gone a long time before we've gotten to AI.

I think it depends on the type of data.

So let's take two examples you kind of brought up there.

So in the mortgage process, the mortgage is, for lack of a better term, a government-mandated commodity.

I mean, at the end of the day, a conforming loan is eventually sold to a bunch of GSEs, and so it has to fit a bunch of standards.

It has to be underwritten exactly the same way.

So all the data collection for a mortgage is every company's collecting the same stuff.

And so in that case,

maybe AI or whatever for more data collection is less interesting on a national or local scale.

But I think your question is also more about, okay, well, you're buying a house in Seattle and there's only so many houses in my price range in the neighborhoods I want to look at in Seattle.

And so what's the value of going after more data?

Well, in that case,

It's actually more data to help the buyer get a better sense because what's unique about real estate is you are making a set of trade-offs.

It's not like you go to Amazon or you're even shopping for a hotel and you actually have to pick from more supply than you could.

You have to make a set of trade-offs when you're buying to say, well, the perfect house is probably never going to come up.

It's like what's available in the time frame I have and the price I have then.

And if you did it 10 times in a row in 10 alternate universes, it would play out differently.

And so when you're making a set of trade-offs,

More data about how to make that trade-off is immensely valuable.

So take a product that Zillow offers called Zillow Showcase, which allows AI tools to generate more of a virtual tour and generate more data to learn about whether that house is good for you before you'd have to go get in a car and waste your Saturday going to the wrong houses.